speech-to-text

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides a command to install the official ElevenLabs CLI by piping a shell script from the vendor's GitHub repository (https://github.com/elevenlabs/cli/releases/latest/download/elevenlabs-cli-installer.sh | sh). Since this originates from the official vendor release channel, it is categorized as safe.
  • [EXTERNAL_DOWNLOADS]: The documentation references and downloads official ElevenLabs packages and tools from trusted sources including GitHub, NPM, and PyPI.
  • [COMMAND_EXECUTION]: The skill suggests running npx -y @elevenlabs/cli@latest onboard init to initialize the project and authenticate the user, which is a standard procedure for using the vendor's services.
  • [INDIRECT_PROMPT_INJECTION]: As a transcription service, the skill ingests and processes audio and video content from user-provided local files or remote URLs (source_url, cloud_storage_url). This is the primary intended use case for the skill.
  • Ingestion points: Media files and URLs passed to transcription methods in the Python and JavaScript SDKs.
  • Boundary markers: The skill documentation does not explicitly detail delimiters for the transcript output, though it is standard for STT tools to return raw text.
  • Capability inventory: The skill has the ability to read local media files and fetch content from the network via the provided ElevenLabs client libraries.
  • Sanitization: The examples show raw transcript text being printed or logged, which is standard behavior for transcription functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 11:59 PM